Anthropic privately pitches its own AI safety bill, Politico reports

Anthropic privately proposed its own frontier AI safety legislation to US lawmakers, Politico reported, citing a document and three people familiar with the discussions.

The proposal would allow a model's release or use to be blocked only for a foreseeable, imminent risk of more than one billion dollars in damage or death or serious injury to more than fifty people. The threat would have to involve biological weapons, loss of developer control or a cyberattack.

The Commerce Department could seek an injunction from the federal district court in Washington after establishing such a threat. Other agencies and officials would be barred from halting development, deployment or internal use.

Politico notes similarities with California AI legislation, including SB 53. Anthropic's proposal is a company draft rather than rules enacted by Congress.